A methane-based time scale for Vostok ice - Maureen E. Raymo
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Quaternary Science Reviews 22 (2003) 141–155
A methane-based time scale for Vostok ice
William F. Ruddimana,*, Maureen E. Raymob
a
Department of Environmental Sciences, University of Virginia, Clark Hall, Charlottesville, VA22903, USA
b
Department of Earth Sciences, Boston University, Boston, MA 02215, USA
Received 7 February 2002; accepted 21 May 2002
Abstract
Tuning the Vostok methane signal to mid-July 301N insolation yields a new ice-core gas time scale. This exercise has two
rationales: (1) evidence supporting Kutzbach’s theory that low-latitude summer insolation in the northern hemisphere controls the
strength of tropical monsoons, and (2) interhemispheric CH4 gradients showing that the main control of orbital-scale CH4
variations is tropical (monsoonal) sources. The immediate basis for tuning CH4 to mid-July insolation is the coincident timing of the
most recent (pre-anthropogenic) CH4 maximum at 11,000–10,500 calendar years ago and the most recent July 301N insolation
maximum (all ages in this paper are in calendar years unless specified as 14C years).
The resulting CH4 gas time scale diverges by as much as 15,000 years from the GT4 gas time scale (Petit et al., Nature 399 (1999)
429) prior to 250,000 years ago, but it matches fairly closely a time scale derived by tuning ice-core d18Oatm to a lagged insolation
signal (Shackleton, Science 289 (2000) 1897). Most offsets between the CH4 and d18Oatm time scales can be explained by assuming
that tropical monsoons and ice sheets alternate in controlling the phase of the d18Oatm signal.
The CH4 time scale provides an estimate of the timing of the Vostok CO2 signal against SPECMAP marine d18O, often used as an
index of global ice volume. On the CH4 time scale, all CO2 responses are highly coherent with SPECMAP d18O at the orbital
periods. CO2 leads d18O by 5000 years at 100,000 years (eccentricity), but the two signals are nearly in-phase at 41,000 years
(obliquity) and 23,000 years (precession). The actual phasing between CO2 and ice volume is difficult to infer because of likely SST
overprints on the SPECMAP d18O signal. CO2 could lead, or be in phase with, ice volume, but is unlikely to lag behind the ice
response.
r 2002 Elsevier Science Ltd. All rights reserved.
0. Introduction between ice-core signals and insolation forcing, as well
as other climatic responses such as ice volume. Accurate
Ice cores contain an incredible variety of climatic phasing (lead/lag) information is vital to efforts to
signals that have had a major impact on climate establish cause-and-effect (forcing-and-response) cli-
research. Over time spans of 400,000 years or more, matic relationships.
ice-core signals show obvious cyclic behavior concen- Section 1 of this paper reviews the history of
trated within the range of astronomically calculated development of time scales for Vostok ice record,
changes in Earth’s orbit, from 20,000 to 100,000 years. including estimates of the uncertainties associated with
To date, however, it has proven difficult to realize the each method. Section 2 sets out the rationale for
full promise of these ice-core signals because the time creating a new Vostok gas time scale based on monsoon
scale has not been sufficiently accurate. Without a forcing of atmospheric methane (CH4) concentrations.
consistently high level of accuracy (to within a few Section 3 makes an initial assessment of this new CH4
thousand years), it is difficult to assess how each ice-core time scale and concludes that it is preferable to the GT4
signal is partitioned among the orbital periods. More (gas) time scale of Petit et al. (1999). Section 4 focuses on
critically, it is impossible to fix the orbital-scale phasing differences between the CH4 time scale and the generally
similar d18Oatm time scale of Shackleton (2000) and
*Corresponding author. Fax: +1-804-982-2137.
concludes that the CH4 time scale provides a more
E-mail addresses: wfr5c@virginia.edu (W.F. Ruddiman), plausible explanation for the small offsets. Section 5 uses
raymo@bu.edu (M.E. Raymo). the CH4 time scale as a basis to define the orbital-scale
0277-3791/03/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 7 7 - 3 7 9 1 ( 0 2 ) 0 0 0 8 2 - 3142 W.F. Ruddiman, M.E. Raymo / Quaternary Science Reviews 22 (2003) 141–155
phasing of CO2 relative to SPECMAP d18O. Finally, ibility record in Southern Ocean core RC11-120, based
Section 6 summarizes factors that make it difficult to on the assumption that the marine magnetic suscept-
infer the phasing of CO2 against ice volume. ibility signal at 411S is a proxy for dust deposition.
Pichon et al. (1992) and Waelbroeck et al. (1995)
correlated deuterium (dD) in Vostok ice with diatom-
1. History of time scales for Vostok ice based estimates of sea surface temperature in marine
cores from middle latitudes of the Southern Ocean based
Early time scales for the Vostok ice record across the on the assumption that mid-latitude changes in sea-
last full interglacial–glacial cycle were derived from ice- surface temperature correlate with air-temperature
flow/ice-accumulation models; these are based on the changes at 801S.
initial rate of accumulation of snow in the area feeding Sowers et al. (1993) tuned the d18O of atmospheric O2
ice to the Vostok core site, and on subsequent ice in Vostok ice to a marine d18O target signal in Pacific
thinning determined from an ice-flow model (Lorius core V19–30. First, they tried to remove the imprint of
et al., 1985). The initial rate of accumulation is bottom-water temperature changes from the benthic
estimated from d18O values in the ice using modern foraminiferal d18O signal to isolate d18Osw (the mean
spatial relationships. The flow model also requires d18O value of seawater) as an ice-volume tuning target.
assumptions about the form of the surface and bedrock Raymo and Horowitz (1996) used an inferred marine
topography upstream from the ice-core site. d13C proxy for past CO2 values to correlate with the
This initial time scale for Vostok ice raised concerns Vostok CO2 record, but also checked matches of Vostok
because the next-to-last deglaciation occurred more than d18Oatm against marine d18O values. Shackleton (2000)
10,000 years before the seemingly equivalent feature in tuned Vostok d18Oatm to a synthetic orbital insolation
the marine d18O signal and was thus out of phase with signal used as a target. He relied on the age of the major
respect to the presumed insolation forcing. This d18Oatm change at the last deglaciation in annually
mismatch was lessened when Jouzel et al. (1993) altered layered GISP ice to set the phase of the precession and
the time scale of Lorius et al. (1985) by incorporating obliquity components of the d18Oatm signal relative to
considerably higher accumulation rates in features insolation.
considered equivalent to marine isotopic stages 6 and 5. Each technique for estimating ages carries inherent
Petit et al. (1999) published signals from the full uncertainties. These errors are difficult to establish for
length of the Vostok ice core. For the upper 110,000 ice-flow/ice-accumulation models, but were estimated by
years of the record, they used a slightly adjusted version Petit et al. (1999) as ranging from 75000 to 15,000 years
of the ice-flow/ice-accumulation time scale from Jouzel in different parts of the Vostok record. Uncertainties in
et al. (1993). They pinned the rest of the ice time scale counting annual layers can be estimated by comparing
(termed ‘‘GT4’’) to levels assumed to represent 110,000 different signals in the same core or the same signal in
and 423,000 years ago, based on inferred correlations to nearby cores (Meese et al., 1997). These range from 100
the marine d18O record. Between these points, they to 200 years back to 12,000 years ago to about 5% of the
estimated ages by calculating ice thinning based on a absolute age down to 55,000 years ago (the level by
flow model, with adjustments for basal melting and which annual layering cannot be reliably detected).
sliding. A second method for dating ice cores is by Techniques that link ice-core signals to marine records
counting annual layers preserved in rapidly deposited produce errors in the correlation process, with progres-
ice. This method cannot be used in the slowly deposited sively larger errors for successively more dissimilar
Vostok ice, but time scales developed in the last 15,000 signals. These techniques also implicitly incorporate
years from Greenland ice can be transferred to Vostok the 3000-year to 5000-year uncertainty in orbital tuning
using globally correlative climate signals such as of the marine record (Imbrie et al., 1984). Correlations
methane. of d18Oatm to marine d18Osw also introduce an un-
A third approach is to correlate (‘‘tune’’) climate- certainty of 7500 years in the turnover time of O2 in the
sensitive proxies within Vostok ice to climatic signals in atmosphere (Sowers et al., 1993). In addition, correla-
marine sediments. In effect, this method transfers the tions between ice and gas phases within Vostok ice may
orbitally tuned time scale of the marine record to the ice be in error by 72000 years (Sowers et al., 1993).
core. Marine records are tuned by matching the filtered
41,000-year and 23,000-year components of marine d18O
signals to assumed forcing at the corresponding orbital 2. Rationale for tuning ice-core methane (CH4) to orbital
periods, with lags appropriate to the long time constants insolation
of ice sheets.
Several such correlations to the marine record have We present here a new gas time scale for Vostok based
been attempted. Petit et al. (1999) correlated the dust on a simple and well-justified approach: tuning the
concentration in Vostok ice with a magnetic suscept- methane (CH4) gas signal in Vostok ice to mid-JulyW.F. Ruddiman, M.E. Raymo / Quaternary Science Reviews 22 (2003) 141–155 143
insolation at 301N. This method is based on two 800
assumptions. The first assumption is that the CH4 signal
is mainly a response to the strength of low-latitude
monsoons, with secondary input from boreal sources
July 30°N Insolation (W/m2)
(Raynaud et al., 1988; Chappellaz et al., 1990; Blunier
700
et al., 1995; Brook et al., 1996). Strong summer 500
monsoons deliver moisture to southeast Asia, India,
CH4 (ppbv)
and northern Africa, and the monsoonal rains fill lakes
and wetlands that become sources of methane as
vegetation decays in reducing conditions. Modeling of 600
CH4 gradients between Greenland and Antarctic ice Y.D.
490
over the last 40,000 years indicates that the tropics have
been the dominant source of methane over orbital time Insolation
scales (Chappellaz et al., 1997; Brook et al., 2000). 500 CH4
The second assumption is that tropical monsoons are
driven by low-latitude summer insolation with a
480
dominant orbital precession signal. This assumption 5.000 10.000 15.000
builds on the orbital monsoon hypothesis first proposed Calender Years BP
by Kutzbach (1981). This hypothesis has since been
supported by a wide array of evidence, including Fig. 1. Comparison of CH4 trends between 15,000 and 5000 years ago
(based on a composite trend from the GRIP ice core assembled by
tropical lake levels (Street-Perrott and Harrison, 1984;
Chappellaz et al., 1997) and July insolation at 301N (from Berger and
Kutzbach and Street-Perrott, 1985), Nile-induced stra- Loutre, 1992). YD marks the Younger Dryas event.
tification of the Mediterranean Sea (Rossignol-Strick
et al., 1982), and influxes of freshwater diatoms blown
from African lakes into Atlantic sediments (Pokras and 21 summer solstice is often used as a reference for
Mix, 1987). analyzing insolation changes, but mid-July insolation
This tuning exercise requires two key choices of an appears to provide the optimal match to the GRIP CH4
insolation ‘‘tuning target’’: (1) a critical latitude, and (2) peak (Fig. 1). GCM analyses of summer monsoon
a critical season or month. To guide this choice, we forcing often choose the June/July/August season,
compare the timing of insolation signals calculated in centered on mid-July, as the critical season (Prell and
calendric years against the calendric age of the most Kutzbach, 1987).
recent CH4 maximum in the GRIP ice core from We chose 301N as the critical insolation latitude to
Greenland as determined from counting annual layers. balance the combined effect of several land masses
The most recent CH4 maximum in the GRIP ice core heated by low-latitude insolation. North Africa lies
prior to the anthropogenic era is a broad maximum almost entirely in the tropics (o231N), but Asia extends
from 14,500 to 8500 years ago (Fig. 1). This peak is through the subtropics and far into higher latitudes.
interrupted by an abrupt CH4 minimum during the Across much of the north tropics and subtropics, a
Younger Dryas event near 12,000 years ago, and is different choice of latitude would not significantly alter
followed by a smaller minimum just before 8000 years the shape (or timing) of the tuning target. But at
ago. Despite these interruptions, the wave-like form of latitudes above 501N, insolation signals begin to take on
the orbital-scale peak remains evident. Peak CH4 values more of the signature of the 41,000-year obliquity signal,
are centered near 11,000 to 10,500 years ago. Attempts and this can nudge specific precessional maxima and
to refine this estimate of the age of the CH4 maximum minima toward slightly younger or older ages.
using a spline fit would be hampered by the presence of This rationale for tuning methane to July 301N
the abrupt Younger Dryas event and the rapid deglacial insolation might be challenged in several ways. One
CH4 increase near 14,500 years ago. At this level in the basis for challenge is that some climatic proxies linked to
ice, the age offset between the CH4 gas and the ice in tropical monsoons reach maximum responses later than
which it is trapped is about 50 years (Blunier et al., the ice-core CH4 maximum. Radiocarbon dating shows
1995). that closed-basin lakes in North Africa reached their
This age for the CH4 maximum matches the timing of highest levels between 9000 and 6000 14C years ago
the most recent mid-July insolation maximum at low (Street-Perrott and Harrison, 1984) and that Nile River
and middle latitudes (Berger and Loutre, 1992). Because runoff from monsoonal regions of northeast Africa
a family of precession curves exists for each season and created organic-rich sapropel muds in the Mediterra-
month due to the slow progression of the solstices and nean Sea from 9000 to 7000 14C years ago (Thunell and
equinoxes around the elliptical orbit, it is necessary to Williams, 1989). Even when converted to calendar years,
select one of these monthly signals as critical. The June these responses occurred 10,000 to 7500 years ago, well144 W.F. Ruddiman, M.E. Raymo / Quaternary Science Reviews 22 (2003) 141–155 after the July insolation maximum from 11,000 to sediments at high latitudes or in shallow marine 10,500 years ago. Similar lags occur in upwelling proxies sediments at any latitude, could be another possible and pollen influxes off the Somali coast of eastern Africa source of atmospheric CH4. The rapid changes in CH4 (Prell, 1984; Prell and van Campo, 1986). These lags, during the last 15,000 years shown in Fig. 1 could be which occur in and around Africa, may reflect external interpreted as lending credibility to this idea. To date, climatic influences on the African monsoon such as the however, ice-core studies have not favored this possibi- lingering effects of northern ice sheets or North Atlantic lity. The character of the record in Fig. 1 is more easily SST (deMenocal and Rind, 1996), or the delayed timing interpreted as one in which cold-climate CH4 minima of the 23,000-year SST maximum in the eastern tropical interrupt longer-term orbital-scale CH4 maxima, rather Atlantic from which the monsoon draws latent heat than as warm-climate pulses of excess CH4 expelled (McIntyre et al., 1989). from clathrates. In addition, Brook et al. (2000) noted On close inspection, however, many of the best-dated that even the most rapid CH4 declines were spread over African lake records resemble the methane trends 200 to 300 years, an interval significantly longer than the plotted in Fig. 1, showing major increases in monsoonal abrupt shifts of other ice-core signals towards colder moisture 14,500 years ago, maximum lake levels by conditions, and consistent with the time scale for 11,000 years ago, and high lake levels until 5000 to 4000 ecosystem adjustments to suddenly imposed reductions years ago (Gasse, 2000). In contrast, the Mediterranean in tropical moisture. sapropels occur well toward the end of this moist In any case, the slowly deposited ice at Vostok acts as interval, suggesting that they are a lagging indicator of a low-pass filter and smoothes most millennial-scale monsoon conditions. CH4 fluctuations, leaving behind a record mainly of In any case, the recent CH4 maximum in the GRIP ice orbital-scale changes. Over this time scale, Chappellaz core used to set the timing of our tuning exercise is well et al. (1997) inferred that any clathrate contribution to dated, and other evidence from Asia supports this ice-core CH4 must have arrived as a relatively steady timing. Planktonic foraminiferal d18O signals in South input over orbital-scale intervals. As such, this input China Sea cores show negative departures from typical would be indistinguishable from monsoonal inputs. d18O values during the last deglaciation (Wang et al., Our method carries less uncertainty than any other 1999). These deviations are interpreted as salinity ice-dating technique. The CH4 signal plotted in Fig. 1 is decreases caused by excess river inflow caused by south in error by at most 150750 years (Meese et al., 1997). Asian monsoons. This inferred monsoon signal first Visual inspection shows that it would be difficult to shift appeared 15,000 years ago, markedly increased in our choice of July as the critical insolation month by strength just before 13,000 years BP, and gradually more than 7300 to 400 years without harming the ended after 9000 years ago. Given the large wetland match shown in Fig. 1. (Such a shift equates to less than expanse in southern Asia, Asian CH4 sources probably a week if converted to equivalent time in the preces- dominated the ice-core CH4 signal. sional cycle.) In addition, monthly (and seasonal) A second possible challenge to our rationale is that insolation signals have no significant dating error across part of the CH4 signal comes from boreal wetland the time range examined here (Berger and Loutre, 1992). sources far from regions controlled by tropical mon- As a result, errors in our choice of tuning target are soons. But CH4 gradients between Greenland and unlikely to exceed 7500 years. Antarctic ice suggest that tropical methane sources were The largest potential source of error in our procedure twice as large as those north of 301N during both the (assuming the validity of our basic rationale) arises from last-glacial CH4 minimum and the late-deglacial CH4 the process of linking maxima and minima in the CH4 maximum (Chappellaz et al., 1997; Brook et al., 2000). signal with the precisely dated insolation signal (Table 1). This indicates that tropical monsoons dominated the The Vostok CH4 signal is sampled at an average interval CH4 signal through a range of climatic changes. of o1500 years over the last 350,000 years, with the In addition, the mid-July insolation peak that spacing ranging from o1000 years in the upper part of produces maximum amounts of CH4 from tropical the record to about 2000 years in the older portion. This monsoons at the 23,000-year cycle should also drive a sample spacing introduces an average uncertainty of coincident CH4 influx from boreal regions. Mid-summer 7750 years across the record, with a range from 7500 heating of the Asian continent should warm the chilly years in the upper part to 71000 years in the older part. northern wetlands, stimulate a brief mid-summer pulse Combined with the 7500-year error in our initial choice of microbial activity, and liberate maximum amounts of of the tuning target, we estimate that the errors boreal CH4. The mid-July phasing of the CH4 maximum associated with our method could optimally be as small in Fig. 1 is consistent with the idea of coincident mid- as 71000–2000 years. summer CH4 influxes from tropical and boreal sources. Across some intervals, however, our method could A third potential criticism is that melting of CH4- have larger errors. Precession-driven CH4 maxima and bearing clathrates, either those frozen in terrestrial minima are not well developed during the glacial parts
W.F. Ruddiman, M.E. Raymo / Quaternary Science Reviews 22 (2003) 141–155 145
Table 1 tion uncertainties occur within low-amplitude CH4
CH4 time scale for Vostok ice oscillations during full-glacial conditions. In addition,
Depth in ice (m) Age (103 yr) Feature correlated some CH4 peaks are interrupted by brief minima, but
the basic precessional wave pattern remains apparent
149.2 0 —
321.2 11 Max through most of the record. We call this time scale the
506.4 22 Min ‘‘CH4 time scale’’ (Table 1).
664.1 32 Max The Vostok record below 3180 m ice depth (350,000
748.3 45 Min years age) is omitted because neither the CH4 signal nor
860.6 58 Max
the 301N insolation tuning target has sufficiently
1087.2 70 Min
1237.2 82 Max distinctive amplitude variations within this age range
1338.2 93 Min to permit credible tuning. It is also unclear whether the
1526.2 104 Max Vostok record reaches the bottom of marine isotopic
1557.4 115 Min stage 11 (Petit et al., 1999).
1893.4 131 Max
2030.9 137 Midpoint
Cross-spectral analysis confirms that the strong
2137.1 149 Max 23,000-year power in the tuned CH4 signal is in phase,
2177.3 161 Min and highly coherent, with precessional insolation
2363.0 174 Max (Fig. 2b–d). The largest mismatches in amplitude occur
2437.0 185 Min on CH4 minima, most of which are rectified at minimum
2525.0 196 Max
values of 400–420 ppb, even when insolation falls to very
2557.7 208 Min
2621.7 219 Max low values. Amplitude mismatches between insolation
2698.0 230 Min and CH4 also occur near terminations, with somewhat
2771.2 240 Max deeper minima just prior to deglaciations and dispro-
2830.4 251 Min portionately strong CH4 maxima early in the following
2857.5 265 Max
interglaciations. These mismatches suggest that factors
2911.4 278 Min
2944.5 290 Max other than insolation also effect the amplitude of early
2994.5 301 Min interglacial CH4 maxima.
3051.5 312 Max The main difference between the two signals is the
3054.5 320 Min concentration of CH4 power at 100,000 and 41,000
3078.5 328 Midpoint
years, periods that are not prominent in the July 301N
3109.5 334 Max
3138.5 338 Midpoint insolation tuning signal (Fig. 2b). The 100,000-year
3165.5 344 Min signal is in phase with eccentricity, largely as a result
of the common eccentricity modulation of both CH4
Feature correlated specifies maxima, minima or transition midpoints in
both the 301N insolation curve and the Vostok CH4 record used as and of the precessional insolation signal used as the
control points for correlation, with linear interpolation of intervening tuning target. The 41,000-year CH4 signal lags well
levels. behind insolation at the obliquity period (Fig. 2c) and is
nearly in phase with the 41,000-year component of
SPECMAP d18O. This phasing suggests that the 41,000-
of the Vostok record, such as the intervals around year CH4 signal largely reflects an overprint of ice
40,000, 150,000, and 260,000 years ago. During these volume (or some other glacial boundary condition) on
intervals, it is difficult to link the CH4 record to methane emissions from tropical and/or boreal wetlands
precessional insolation. As a result, errors larger than (Prell and Kutzbach, 1987).
5000 years (1/4 of a precessional cycle) are not The methane record plotted at the CH4 time scale is
impossible. For the rest of the record, errors greater compared with that of the GT4 time scale of Petit et al.
than 2000 years are unlikely. Finally, as is the case for all (1999) in Fig. 3. Because the CH4 time scale used CH4
tuning techniques, our method is vulnerable to the gas for tuning, we use the GT4 gas time scale for this
assumption that the phase of the monsoon may have comparison, rather than the GT4 time scale for ice. The
varied through time because of variable influences of GT4 gas time scale was derived by subtracting 2000 to
other parts of the climate system (ice sheets, SST, etc). 6000 years from the time scale for ice, with larger
differences during glacial intervals of slow ice accumula-
tion (Petit et al. 1999).
3. Initial assessment of the CH4 time scale For most of the record back to 250,000 years ago, the
CH4 and GT4 time scales are similar, with age
A time-series comparison shows the tuned correlation departures of no more than 5000 to 10,000 years, and
between Vostok CH4 and 301N July insolation (Fig. 2a). generally only a few thousand years (Fig. 3a). The ages
Both signals show distinct amplitude modulation at the of many CH4 maxima and minima in the GT4 time scale
precessional cycle. As noted above, the major correla- only needed to be shifted by 1000 to 3000 years to create146 W.F. Ruddiman, M.E. Raymo / Quaternary Science Reviews 22 (2003) 141–155
800
July 30N
Vostok CH4
Vostok methane (ppbv)
July insolation (W/m2)
520
600
480
400
440
200
(a) 0 50 100 150 200 250 300 350
100 kyr
41 kyr
23 kyr
5 104 2 105
July 30N
4 104 Vostok CH4
1.5 105
Rel. power
3 104
BW
1 105
4
2 10
5 104
1 104
(b) 1.0 0 100
0.8 95% CL
Coherency
0.6
0.4
0.2
(c) 0.0
90
Phase (deg.)
0
-90
0.00 0.05 0.10
(d) Frequency kyr -1
Fig. 2. Comparison of Vostok CH4 (plotted on the CH4 time scale) and July 301N insolation for the last 350,000 years: (a) Time series; (b–d) Cross-
spectral analysis; (b) power spectra; (c) coherence between CH4 and insolation; and (d) relative phasing at orbital periods. Spectral analysis followed
Blackman–Tukey method, with 1/3 lags. Phases at the major orbital periods shown with 95% confidence intervals.
the CH4 time scale. These age offsets lie well within the suspicious-looking distortions of other (untuned) gas
uncertainties estimated by Petit et al. (1999). signals from Vostok ice (Sections 4 and 5).
The CH4 time scale produces several relatively abrupt Prior to 250,000 years ago, the ages derived from the
changes in ice accumulation rate compared to the GT4 CH4 time scale diverge from those of the GT4 time scale
time scale. Although these might result from ‘‘over- by much larger amounts averaging near 15,000 years
tuning’’ the Vostok methane record to the July 301N (Fig. 3a). Over much of this interval, the GT4 time scale
insolation target, none of these changes produces yields a climatically implausible phasing betweenW.F. Ruddiman, M.E. Raymo / Quaternary Science Reviews 22 (2003) 141–155 147
350
Age CH4 (this study)
300 Tuned Age (Shackleton et al., 2000)
1:1 line
250
Tuned gas ages (kyr) 200
150
100
50
0
0 50 100 150 200 250 300 350
(a) Glaciological (GT4) gas age (kyr)
800 600
Age GT4 (gas)
Vostok methane (ppbv)
July insolation (W/m2)
July 30 N
600
500
400
400
0 50 100 150 200 250 300 350
(b) Age (kyr)
18
Fig. 3. Comparison of GT4, CH4 and d Oatm time scales: (a) Diagonal (1:1) line based on Vostok GT4 gas time scale (Petit et al., 1999) is used as a
reference for departures of ages in other time scales. Solid line: CH4 time scale from this paper; dotted line: d18Oatm time scale of Shackleton (2000);
(b) Time series of Vostok CH4 on the GT4 time scale against July insolation.
Vostok CH4 values and summer (July 301N) insolation phasing of orbital-scale ice-core signals. It also suggests
(Fig. 3b): maxima in CH4 are often aligned with minima that it would be worthwhile to take a closer look at the
in July insolation, a phasing opposite what would be differences between these two time scales.
expected for northern hemisphere insolation control of
monsoonal and boreal CH4 input.
Also plotted in Fig. 3a are age-depth offsets between
the d18Oatm time scale of Shackleton (2000) and the GT4 4. Assessment of the d18Oatm Signal
time scale. For levels prior to 250,000 years ago, the
d18Oatm and CH4 time scales agree in showing ages The d18Oatm time scale is based on tuning ice-core
18
considerably older than those from GT4. This conver- d Oatm to a synthetic insolation signal containing both
gence supports the conclusion that the GT4 age precession and obliquity, and incorporating phase lags
estimates for this older part of the Vostok record are appropriate to the slow responses of ice sheets to orbital
substantially in error. forcing at these periods. In contrast, as summarized in
The fact that the CH4 and d18Oatm time scales Section 2, the CH4 time scale is based on tuning CH4 to
generally differ from each other by only a few thousand a precession-dominated 301N July insolation with no
years points to major progress toward the goal of lags. Although the d18Oatm and CH4 time scales are
obtaining a Vostok time scale sufficient to analyze the based on different tuning rationales, they differ in age by148 W.F. Ruddiman, M.E. Raymo / Quaternary Science Reviews 22 (2003) 141–155
less than 5000 years through most of the last 350,000 phase of the d18Oatm signal through time. We compare
years, and by an average of less than 500 years. the d18Oatm signal (plotted on the CH4 time scale) to
Despite this overall agreement, larger offsets between mid-July insolation at 301N (Fig. 4b) and to the
the two time scales do occur in some intervals (Fig. 3a). SPECMAP d18O curve (Fig. 4c).
One clue to the origin of these offsets is that the relative The d18Oatm trend shows fairly systematic phase
phasing of the CH4 and d18Oatm signals varies through offsets with respect to both insolation and SPECMAP
time, with the two signals sometimes closely in phase, d18O (as also noted by Jouzel et al., 2002). During peak
and sometimes well out of phase (Fig. 4a). These shifting interglacial substages (marine isotopic stages 5.5, 9.3,
offsets cannot be an artifact of the time scale used: both and, to some extent, 7.3), the d18Oatm signal is
CH4 and d18Oatm are recorded in the same gas phase of approximately in phase with July insolation (Fig. 4b),
the same ice core, and so similar offsets must exist for but it generally lags behind insolation during late-
any time scale chosen. interglacial substages (5.1, 5.3, 7.1, and 9.1) and glacial
An initial way to explore the cause of these offsets is isotopic stages (4–2, 6 and 8). Conversely, d18Oatm leads
to assume that the CH4 time scale correctly fixes the age SPECMAP d18O during and just after peak-interglacial
of Vostok ice, and then examine resulting changes in the substages (Fig. 4c), but is roughly in phase during most
800
-0.4
Vostok atmospheric δ18O (‰)
0.0
Vostok CH4 (ppbv)
600
0.4
0.8
400
1.2 δ Oatm
18 CH 4
(a)
-0.4 5.1
5.5 7.1 7.3
5.3
7.5 9.3
9.1 520
Vostok atmospheric δ18O (‰)
0.0 July insolation 30 N (W/m2)
0.4 480
0.8
440
1.2 δ18 Oatm July 30N insolation
(b)
5.5
-0.4 5.1
9.3 2
5.3 7.1 7.3 9.1
Vostok atmospheric δ18O (‰)
7.5
SPECMAP δ18O stack (‰)
0.0
1
0.4
0
0.8
-1
1.2 δ18 O atm SPECMAP δ18 O stack
-2
0 50 100 150 200 250 300 350
(c) Age (kyr)
18
Fig. 4. Evaluation of the Vostok d Oatm signal by time-series comparison with: (a) Vostok CH4 (both records plotted on the CH4 time scale);
(b) July 301N insolation (d18Oatm plotted on the CH4 time scale and July 301N insolation from Berger and Loutre (1992); (c) d18Oatm plotted on the
CH4 time scale and SPECMAP d18O plotted on the time scale of Imbrie et al. (1984).W.F. Ruddiman, M.E. Raymo / Quaternary Science Reviews 22 (2003) 141–155 149 late-interglacial and glacial isotopic stages. These shifts inferred that the largest (estimated) Dole effect occurred in d18Oatm phasing appear to be linked to climate in a between peak-interglacial substages 5.5 and 5.4, ac- systematic way, but why? counting for 50% or more of the d18Oatm change Two climatic factors control d18Oatm variations over (monsoon control). The smallest Dole effect occurs in orbital time scales. The d18O value of oxygen in the late-interglacial substages and glacial stages, where it atmosphere (d18Oatm) responds to the globally averaged accounts for as little as 20 to 30% of the total (ice-sheet d18O value of marine sea water (d18Osw), with an control). estimated lag today of about 1200 years caused by the We began this section by making the assumption that turnover time of O2 in the atmosphere (Bender et al., the ice-core methane signal has the constant phasing 1994). Because d18Osw is a measure of the size of global with July insolation forcing used in creating the CH4 ice sheets (and excludes the effect of ocean temperature time scale. Now we test the opposite assumption: could on regional marine d18O measurements), one of the two the phasing of d18Oatm be constant and that of CH4 major climatic controls on d18Oatm is thus global ice variable? volume. The tuning strategy used by Shackleton (2000) gave The second control operates through the ‘‘Dole the d18Oatm signal the same phasing as SPECMAP used effect’’ (Dole et al., 1954). Fractionation during marine for the marine d18O signal: a 5000-year lag of d18O and terrestrial photosynthesis and respiration causes behind June 21 precessional insolation and an 8000-year modern d18Oatm to be offset by +23.5% compared to lag behind obliquity. He chose smaller marine d18O lags d18Osw. Sowers et al. (1993) found that the estimated size behind insolation than those in SPECMAP, but also of this effect varied over orbital time scales by about incorporated the additional lag of d18Oatm behind 1%, roughly the same magnitude as changes in d18Osw marine d18O. Because Shackleton tuned the d18Oatm caused by ice sheets. Bender et al. (1994) argued that signal primarily to orbital precession, the average lag of these changes in the Dole effect probably resulted from the d18Oatm signal behind insolation was about 5000 varying photosynthesis and respiration in the tropical years. monsoon system. Thus, changes in tropical monsoons If we accept Shackleton’s choice of d18Oatm phasing as are the second likely climatic control on d18Oatm. correct, how would it affect the timing of the CH4 We propose that the variable timing of the d18Oatm signal? During and just after peak interglaciations, when signal relative to insolation and SPECMAP d18O d18Oatm and CH4 are in phase, CH4 should share the (Fig. 4) is caused primarily by changes in the relative same 5000-year lag behind June 21 insolation as the influence of these two controls on d18Oatm through time. d18Oatm signal. For the remainder of each interglacial- During and just after peak-interglacial stages, the glacial cycle, when CH4 leads d18Oatm by about 5000 d18Oatm signal responds in phase with July insolation years, CH4 should be approximately in phase with June and CH4 because the monsoon system controls d18Oatm 21 insolation. in the absence of large ice sheets. At these times, the But this pattern of variable CH4 phasing makes little d18Oatm signal joins CH4 in following the fast response physical sense in climatic terms. Why would a monsoon- to insolation forcing and minimal lag typical of dominated CH4 signal respond earlier when ice sheets monsoons. In addition, the rapid and early ice melting are large, but later when ice sheets are small? If ice sheets typical of terminations brings the d18Osw component of have any effect on monsoon timing, it should be to the d18Oatm signal into close alignment with the retard them. We conclude that the differences between monsoonal Dole component. the CH4 and d18Oatm time scales are more plausibly At other times, including the later parts of interglacial explained by variable controls on the phasing of the stages and the glacial stages, the d18Oatm signal responds d18Oatm signal than by variable timing of the CH4 signal. with a phase closer to that of marine d18Osw (ice What about the possibility that ice sheets can indeed volume), because the monsoon weakens from its peak- cause delays in monsoon timing? If growing ice sheets interglacial strength and because ice-sheet variations do retard monsoons, then the phase of the CH4 signal grow large enough to become the major control of the should shift to younger ages during glacial isotopic d18Oatm signal. At these times, the d18Oatm signal takes stages because of the larger lags behind insolation on the slow response to insolation changes typical of ice forcing. And if the gas-phase CH4 signal has been sheets, and it lags behind the monsoon-controlled CH4 shifted in this way, the d18Oatm signal must also shift to signal. In summary, the phasing of the d18Oatm signal is younger ages because it is recorded in the same bubbles. monsoon-dominated when ice sheets are small, and ice- But such a shift in ice-core d18Oatm creates a major dominated when ice sheets are larger. problem for its phase relationship with marine d18O Other evidence supports this interpretation. The size (Fig. 4c). Any major shift of the ice-core d18Oatm signal of the monsoonal Dole effect can be monitored by to younger ages will make it lag marine d18O at most variations in amplitude of the d18Oatm signal in excess of isotopic transitions. If this occurs, an implausible those in the marine d18Osw signal. Sowers et al. (1993) scenario is created: the ice-core d18Oatm signal now lags
150 W.F. Ruddiman, M.E. Raymo / Quaternary Science Reviews 22 (2003) 141–155
well behind its only known controlling processes: the prominent isotopic transitions occurring nearly in
monsoon signal (through the Dole effect), and the phase. The largest exception is a CO2 lead of 4000 to
marine d18O signal. Because this scenario makes no 5000 years at the isotope stage 6/5 deglaciation
sense, it seems unlikely that ice sheets can cause major (termination II). In contrast, the CO2 signal plotted on
lags in the monsoon-generated CH4 signal. the GT4 time scale shows a more erratic pattern of leads
Although it is difficult (impossible?) to show conclu- and lags compared to SPECMAP d18O (Fig. 6a).
sively that the monsoon-dominated CH4 signal retains This improved correlation is confirmed in a quantita-
the same phasing through all the complex changes tive way by cross-spectral analysis (Figs. 5b–d and 6b-d).
occurring during glacial–interglacial climate cycles, the The 23,000-year orbital period has a greatly improved
evidence reviewed above suggests that it is likely that the coherency in the CH4 time scale, and the already strong
CH4 signal retains a stable (or very nearly stable) phase coherence of the 41,000-year and 100,000-year cycles in
through time. In summary, the time scales derived by the GT4 time scale both increase in the CH4 time scale.
tuning ice-core d18Oatm and CH4 to insolation diverge Because CO2 was not used to tune the CH4 time scale,
from the time scale of Petit et al. (1999) prior to 250,000 the improved coherences at all three orbital periods are
years ago, but generally match each other to within a independent confirmation of the validity of the CH4
few thousand years back to 350,000 years ago. On time scale. In comparison, the d18Oatm time scale from
average, the CH4 time scale is just 450 years younger Shackleton (2000) has power at 23,000 years, but no
than the d18Oatm time scale of Shackleton (2000). If the distinct peak at that period.
smaller differences between the two tuned time scales are The more similar phasing of the CO2 and d18O signals
indeed caused mainly by variable climatic controls on resulting from the CH4 (and d18Oatm) time scale fits into
d18Oatm phasing, the CH4 time scale is the more accurate a long trend in the history of Vostok time-scale
of the two. This level of accuracy (to within a few development. Earlier time scales based on models of
thousand years) allows a close examination of the ice flow and accumulation (Lorius et al., 1985), and
phasing of other gases in Vostok ice. adjusted from ice age to gas age (Barnola et al., 1987),
produced an irregular phasing between CO2 and marine
d18O: large CO2 leads occurred on terminations, but
5. Comparison of Vostok CO2 Versus SPECMAP d18O sizeable lags occurred at the marine isotopic stages 5/4
and 5e/5d transitions (times of inferred major ice-sheet
Of all the signals recorded in Vostok ice, the growth). Later time scales (summarized in Section 2)
concentration of carbon dioxide is the most important generally tended to reduce this variability in phasing,
to global climate. Because the CO2 signal was not used and the CH4 time scale devised here points to a still
in the tuning process, it provides an independent way to more consistent (in-phase) timing.
evaluate the CH4 time scale, in this case by comparison As noted earlier, no a priori basis exists for predicting
to the GT4 time scale of Petit et al. (1999). For cross- the form of the CO2 signal. Still, it makes good sense
spectral comparisons, we omit levels above 5000 years from a climate-theory viewpoint that the phasing
because of likely human influences on CH4 (Ruddiman between CO2 and ice volume should be consistent rather
and Thomson, 2001). than irregular. The relatively stable (and nearly in-
The concentration of spectral power for CO2 at phase) relationship across major d18O transitions in the
orbital periods is not much different for the CH4 and CH4 time scale is suggestive of a consistent physical
GT4 time scales (Figs. 5 and 6). In the CH4 time scale, relationship between CO2, a major greenhouse gas, and
the peaks at 100,000 and 41,000 years lose power d18O, a primary indicator of climate. The increases in
compared to GT4, but a new peak emerges near 23,000 coherency at all three orbital periods provide further
years. Also, compared to the d18Oatm time scale of evidence of a more consistent physical link between CO2
Shackleton (2000), the CH4 time scale has almost and SPECMAP d18O over the last 350,000 years.
identical amounts of power at 100,000 and 41,000 years, For the CH4 time scale, maximum coherencies occur
but a substantially larger peak at 23,000 years. Because at a CO2 lead of 121 (140071400 years) relative to
no a priori expectation exists about the ‘‘true’’ shape of SPECMAP d18O for the 41,000-year period and 311
the CO2 signal, and the distribution of its power among (20007900 years) for the 23,000-year period. These
the orbital periods, however, these comparisons of CO2 phases for the d18Oatm time scale agree to within 50–100
spectra alone are not a revealing test of the CH4 time years. Because changes at these two periods produce
scale. much of the distinctive character of rapid transitions in
A more diagnostic evaluation comes from comparing the SPECMAP d18O signal (Imbrie et al., 1992), the
the relative timing of Vostok CO2 and the SPECMAP general visual similarity of the CO2 and d18O trends
marine d18O signal (Imbrie et al., 1984). Plotted on the results from the similar phasing at these periods. At the
CH4 time scale (Fig. 5a), the CO2 signal is visually 100,000-year period, CO2 leads SPECMAP d18O by 231
similar to SPECMAP d18O, with many of the most (660072000 years).W.F. Ruddiman, M.E. Raymo / Quaternary Science Reviews 22 (2003) 141–155 151
CO2 (ppmv) SPECMAP δ18 O 2
280
SPECMAP δ18O stack (‰)
Vostok CO2 (ppmv)
1
240
0
-1
200
-2
0 50 100 150 200 250 300 350
(a) Age (kyr)
100 kyr
41 kyr
23 kyr
4 10 4 6 101
CO2
SPECMAP δ18 O
Rel. power
BW
2 10 4 3 101
(b) 0 10 0 0 100
0.8 95% CL
Coherency
0.6
0.4
0.2
(c) 0.0
CO2 lead
90
Phase (deg.)
0
-90
CO2 lag
0.00 0.05 0.10
(d) Frequency kyr-1
Fig. 5. Comparison of Vostok CO2 (on the CH4 time scale) and SPECMAP d18O (on the time scale of Imbrie et al., 1984). (a) Time series. (b–d)
Cross-spectral analysis: (b) power spectra; (c) coherency between CH4 and d18O signals; and (d) relative phasing at orbital periods. Spectral analysis
followed Blackman–Tukey method, with 1/3 lags. Phases at the major orbital periods shown with 95% confidence intervals.
6. Discussion: Climatic role of CO2 in ice-volume changes explore this issue. On the CH4 time scale, Vostok CO2
leads SPECMAP d18O at all three orbital periods
One of the key issues in studies of orbital-scale climate (Fig. 5). Before this can be taken as evidence that CO2
change is the phasing between CO2 and ice volume. forces ice volume, however, two problems must be
Accurate definition of the relative phasing of these two addressed.
signals should reveal critical cause-and-effect links in the The first problem is that, based on a wide array of
climate system. The evidence above indicates that the evidence, the SPECMAP time scale is probably in error,
CH4 time scale for Vostok ice is sufficiently accurate to giving ages for the d18O signal that average about 1500152 W.F. Ruddiman, M.E. Raymo / Quaternary Science Reviews 22 (2003) 141–155
CO2 (ppmv)) SPECMAP δ18 O 2
280
SPECMAP δ18O stack (‰)
Vostok CO2 (ppmv)
1
240
0
-1
200
-2
0 50 100 150 200 250 300 350
(a) Age (kyr)
100 kyr
41 kyr
23 kyr
4 10 4 6 101
CO2
SPECMAP δ18 O
Rel. power
BW
2 10 4 3 101
(b) 0 10 0 0 100
0.8 95% CL
Coherenc y
0.6
0.4
0.2
(c) 0.0
CO2 lead
90
Phase (deg.)
0
-90
CO2 lag
0.00 0.05 0.10
(d) Frequency kyr-1
Fig. 6. Comparison of Vostok CO2 (on the GT4 time scale) and SPECMAP d18O (on the time scale of Imbrie et al., 1984): (a) Time series; (b and d)
Cross-spectral analysis: (b) power spectra; (c) coherency between CH4 and d18O signals; and (d) relative phasing at orbital periods. Spectral analysis
followed Blackman–Tukey method, with 1/3 lag. Phases at the major orbital periods shown with 95% confidence intervals.
years too young. Well-dated coral reefs provide a direct A second line of evidence emerged from an attempt
index of past ice-volume changes at some key isotopic to tune the d18O signal using a different rationale.
transitions. By dating reefs formed at the end of Pisias et al. (1990) allowed the lags of the 41,000-year
termination II and during other high stands of sea level, and 23,000-year d18O responses behind the insolation
Edwards et al. (1987) and Bard et al. (1990) found forcing signals to vary through time in order to
evidence for faster deglacial sea-level rises (ice-volume optimize matches of the resulting d18O signal to
decreases) than implied by the SPECMAP d18O curve. those observed in marine records. This approach
Both concluded that coral-reef data indicate deglacia- contrasts with the constant phase lags assumed by
tion some 3000 years earlier than the timing of the SPECMAP in the tuning process (Imbrie et al., 1984).
SPECMAP d18O signal across termination II. Pisias et al. (1990) found that on average the d18O signalW.F. Ruddiman, M.E. Raymo / Quaternary Science Reviews 22 (2003) 141–155 153 was shifted earlier by 1500 years compared to the necessarily indicate the phasing of CO2 with respect to SPECMAP time scale. The largest offset from SPEC- true ice volume. MAP, 3500 years on termination II, agrees with the Shackleton (2000) challenged the assumption that the coral-reef dates. 100,000-year component of d18O signals is an accurate A third line of evidence comes from an indirect record of ice volume. He found that half of the 100,000- comparison of the CH4 time scale to the SPECMAP year d18O signal in Pacific Ocean benthic foraminifera d18O time scale using the d18Oatm time scale of could be explained by changes in Pacific deep-water Shackleton (2000) as an intermediary. Shackleton temperature, and the other half by ice volume. Within (2000) reported that his d18Oatm time scale was on the sizeable errors of his multi-step analysis, the observed average 2000 years older (earlier) than that of SPEC- phase of the 100,000-year d18O signal in this core turned MAP, whereas we find that the CH4 time scale averages out to be partitioned into an early phased temperature 450 years younger than his d18Oatm time scale. Combin- component and a late ice-volume component. ing these results, the CH4 time scale presented here is The same problem complicates our attempt to consistent with a 1550-year shift of the SPECMAP d18O compare the phasing of CO2 against ice volume based time scale toward older (earlier) ages, as suggested by on the SPECMAP d18O record. The average glacial- Pisias et al. (1990). interglacial d18O amplitude of the cores used in the An array of evidence thus suggests that the SPEC- SPECMAP stack is 1.7% (Imbrie et al., 1984). If 0.9– MAP d18O signal lags true ice volume by as much as 1.3% of this total is caused by changes in ice volume 3000 years or more during rapid deglaciations and by an (Schrag et al., 1996; Shackleton, 1967), a residual signal average of about 1500 years throughout the record. of between 0.4% and 0.8% remains, equivalent to an With the SPECMAP time scale adjusted by 1500 years SST change of 1.7–3.31C. Independent estimates of SST toward older ages, the 1400-year CO2 lead relative to changes in the tropical and Southern Ocean cores from SPECMAP d18O (and inferred ice volume) at the which the SPECMAP d18O stack was compiled are 41,000-year period disappears entirely, as does much consistent with a temperature overprint of this size. of the 2000-year lead at the 23,000-year period. CO2 Because an SST overprint of 0.4–0.7% is larger than retains a lead of 5000 years at the 100,000-year period of either the 23,000-year or 41,000-year components of the orbital eccentricity. SPECMAP d18O signal, and amounts to half or more of If CO2 is nearly in phase with d18O (ice volume) at a the filtered 100,000-year d18O component (Imbrie et al., given orbital period, it must be acting mainly as a 1989), it could have a major effect on the phase of any or positive (amplifying) feedback to the ice sheets, rather all of these d18O signals. SPECMAP concluded that SST than as an independent source of forcing. This conclu- leads d18O at all three orbital periods in the tropical and sion derives from the long time constant of ice-sheet Southern Ocean region of the SPECMAP cores (Imbrie responses (Weertman; 1964; Imbrie et al., 1984). Ice et al., 1992, 1993). As a result, any SST overprint in this volume should lag thousands of years behind its primary region would probably be phased ahead of the d18O source(s) of forcing. Ice sheets cannot be strongly forced signals observed. If this is the case, the phase of the true by a CO2 signal that has nearly the same phasing, except ice-volume signal would have to lag behind d18O to in the sense of receiving a self-amplifying positive balance the early SST overprint. This means that CO2 is feedback. The close phase relationship between CO2 likely to lead ice volume by at least as much as it leads and d18O at the 41,000-year obliquity cycle is consistent d18O, and possibly by more. with this kind of feedback relationship. In contrast, the Without a much more detailed analysis, the over- CO2 leads versus d18O at the 100,000-year cycles (and to printing effect of SST on d18O at each orbital period a lesser extent the 23,000-year cycle) still permit an cannot be deciphered. We can rule out the possibility active forcing role for CO2, if d18O is an accurate index that CO2 lags behind ice volume at any of the orbital of ice volume. periods. But we cannot distinguish between the possi- Unfortunately, marine d18O signals are not exact bility that CO2 leads ice volume, in which case it is an proxies for ice volume. The major complication is that active part of the forcing of the ice sheets, or is in phase marine d18O records also contain local temperature with ice volume, in which case it plays the role of an signals. Estimates of the fraction of marine d18O signals amplifying positive feedback to ice-volume changes. caused by temperature changes range from 1/3 (Shack- leton, 1967) to 1/2 (Schrag et al., 1996), and the effect of the temperature overprints on d18O phasing is unclear Acknowledgements (Mix et al., 2001). A second complication is that changes in d18O composition of ice sheets through time may M.E. Raymo thanks NSF for the research support produce offsets between marine d18O and ice volume provided by the MGG and ODP panels and specifically (Mix and Ruddiman, 1984). As a result, the phasing of a generous supplement to grant OCE-9631759. CO2 with respect to the marine d18O signal does not W. Ruddiman thanks his slowly declining TIAA/CREF
154 W.F. Ruddiman, M.E. Raymo / Quaternary Science Reviews 22 (2003) 141–155
investments for financial support. We also thank Alan Yiou, F., Yiou, P., 1993. Extending the Vostok ice-core record of
Mix and an anonymous reviewer for careful and paleoclimate to the penultimate glacial period. Nature 364, 407–412.
comprehensive reviews of the initial manuscript. Jouzel, J., Hoffman, G., Parrenin, F., Waelbroeck, C., 2002.
Atmospheric 18O and sea level. Quaternary Science Reviews 21,
307–314.
References Kutzbach, J.E., 1981. Monsoon climate of the early Holocene: climate
experiment with earth’s orbital parameters for 9000 years ago.
Bard, E., Hamelin, B., Fairbanks, R.F., 1990. U–Th ages obtained by Science 214, 59–61.
mass spectrometry in corals from Barbados: sea level during the Kutzbach, J.E., Street-Perrott, F.A., 1985. Milankovitch forcing of
past 130,000 years. Nature 346, 456–458. fluctuations in the level of tropical lakes from 18 to 0 kyr BP.
Barnola, J.M., Raynaud, D., Korotkevitch, Y.S., Lorius, C., 1987. Nature 317, 130–134.
Vostok ice cores provides 160,000-year record of atmospheric CO2. Lorius, C., Jouzel, J., Ritz, C., Merlivat, L., Barkov, N.I.,
Nature 329, 408–414. Korotkevtich, Y.S., Kotlyakov, V.M., 1985. A 150,000-year
Bender, M., Sowers, T., Labeyrie, L., 1994. The Dole effect and its climatic record from Antarctic ice. Nature 316, 591–596.
variations during the last 130,000 years as measured in the Vostok Meese, D.A., Gow, A.J., Alley, R.B., Zielinski, G.A., Grootes, P.M.,
ice core. Global Biogeochemical Cycles 8, 363–376. Ram, K.C., Taylor, K.C., Mayewski, P.A., Bolzan, J.F., 1997. The
Berger, A., Loutre, M.F., 1992. Astronomical solutions for paleocli- Greenland ice sheet project 2 depth-age scale: methods and results.
mate studies over the last 3 million years. Earth and Planetary Journal Geophysical Research 102, 26411–26423.
Science letters 111, 369–392. McIntyre, A., Ruddiman, W.F., Karlin, K., Mix, A.C., 1989. Surface
Blunier, T., Chappellaz, J., Schwander, J., Stauffer, J., Raynaud, D., water response of the equatorial Atlantic Ocean to orbital forcing.
1995. Variations in atmospheric methane concentration during the Paleoceanography 4, 19–55.
Holocene epoch. Nature 374, 46–49. Mix, A.C., Ruddiman, W.F., 1984. Oxygen-isotope analyses and
Brook, E.J., Sowers, T., Orchado, J., 1996. Rapid variations in Pleistocene ice volume. Quaternary Research 21, 1–20.
atmospheric methane concentration during the past 110,000 years. Mix, A., Bard, E., Schneider, R., 2001. Environmental processes of the
Science 273, 1087–1091. ice age: land, oceans, glaciers (EPILOG). Quaternary Science
Brook, E.J., Harder, S., Severinghaus, J., Steig, E., Sucker, C.M., Reviews 20, 627–657.
2000. On the origin and timing of rapid changes in atmospheric Petit, J.R., Jouzel, J., Raynaud, D., Barkov, N.I., Barnola, J.-M.,
methane during the last glacial period. Global Biogeochemical
Basile, I., Bender, M., Chappellaz, J., Davis, M., Delaygue, G.,
Cycles 14, 559–572.
Delmotte, M., Kotlyakov, V.M., Lipenkov, V., Lorius, C., Pepin,
Chappellaz, J., Barnola, J.-M., Raynaud, D., Korotkevitch, Y.S.,
L., Ritz, C., Saltzman, E., Stievenard, M., 1999. Climate and
Lorius, C., 1990. Atmospheric CH4 record over the last climatic
atmospheric history of the last 420,000 years from the Vostok ice
cycle revealed by the Vostok ice core. Nature 345, 127–131.
core, Antarctica. Nature 399, 429–436.
Chappellaz, J., Blunier, T., Kints, S., Dallenbach, A., Barnola, J.-M.,
Pichon, J.-J., Labeyrie, L.D., Bareille, G., Labracherie, M., Duprat, J.,
Schwander, J., Raynaud, D., Stauffer, B., 1997. Change in the
Jouzel, J., 1992. Surface water temperature changes in the high
atmospheric CH4 gradient between Greenland and Antarctica
latitudes of the southern hemisphere over the last glacial–
during the Holocene. Journal of Geophysical Research 102 (D),
interglacial cycle. Paleoceanography 7, 289–318.
15987–15997.
Pisias, N.G., Mix, A.C., Zahn, R., 1990. Nonlinear response in the
deMenocal, P.B., Rind, D., 1996. Sensitivity of subtropical African
global climate system: evidence from benthic oxygen isotopic
and Asian climate to prescribed boundary condition changes:
record in core RC13-110. Paleoceanography 5, 147–160.
model implications for the Plio–Pleistocene evolution of low-
Pokras, E.M., Mix, A.C., 1987. Earth’s precession cycle and Quaternary
latitude climate. In: Johnson, T.C., Okada, E.O (Eds.), The
Limnology, Climatology, and Paleoclimatology of the East African climatic change in tropical Africa. Nature 326, 486–487.
Lakes. Gordon and Breach, Amsterdam, pp. 57–77. Prell, W.L., 1984. Monsoonal climate of the Arabian Sea during the
Dole, M., Lane, G.A., Rudd, D.P., Zuakelies, D.A., 1954. Isotopic late Quaternary: a response to changing solar radiation. In: Berger,
composition of atmospheric oxygen and nitrogen. Geochimica A.L., et al. (Ed.), Milankovitch and Climate. Reidel, Dordrecht,
Cosmochimica Acta 6, 65–78. pp. 349–366.
Edwards, R.L., Chen, J.H., Ku, T.-L., Wasserburg, G.J., 1987. Precise Prell, W.L., van Campo, E., 1986. Coherent response of Arabian Sea
timing of the last interglacial period from mass-spectrometric upwelling and pollen transport to late quaternary monsoonal
determination of 230Th in corals. Science 236, 1547–1553. winds. Nature 323, 526–528.
Gasse, F., 2000. Hydrological changes in the African tropics since the Prell, W.L., Kutzbach, J.E., 1987. Monsoon variability over the last
last glacial maximum. Quaternary Science Reviews 19, 189–211. 150,000 years. Journal Geophysical Research 92, 8411–8425.
Imbrie, J., et al., 1984. The orbital theory of Pleistocene climate: Raymo, M.E., Horowitz, M., 1996. Organic carbon paleo-pCO2 and
support from a revised chronology of the marine d18O record. In: marine-ice core correlations. Geophysical Research Letters 23,
Berger, A.L., et al. (Ed.), Milankovitch and Climate, Part I. Riedel, 367–370.
Dordrecht, pp. 269–305. Raynaud, D., Chappellaz, J., Barnola, J.M., Korotkevitch, Y.S.,
Imbrie, J., McIntyre, A., Mix, A., 1989. Oceanic response to orbital Lorius, C., 1988. Climatic and CH4 cycle implications of glacial-
forcing in the late Quaternary: observational and experimental interglacial CH4 change in the Vostok ice core. Nature 333,
strategies. In: Berger, A., et al. (Ed.), Climate and Geo-Sciences. 655–657.
Kluwer Academic Publications, Dordrecht, pp. 121–164. Rossignol-Strick, M., Nesteroff, W., Olive, P., Vergnaud-Grazzini, C.,
Imbrie, J., et al., 1992. On the structure and origin of major glaciation 1982. After the deluge: Mediterranean stagnation and sapropel
cycles. I. Linear responses to Milankovitch forcing. Paleoceano- formation. Nature 295, 105–110.
graphy 7, 701–738. Ruddiman, W.F., Thomson, J.S., 2001. The case for human causes of
Imbrie, J., et al., 1993. On the structure and origin of major glaciation increased atmospheric CH4 over the last 5000 years. Quaternary
cycles. II. The 100,000-year cycle. Paleoceanography 8, 699–735. Science Reviews 20, 1769–1777.
Jouzel, J., Barkov, N.I., Barnola, J.M., Bender, M., Chappellaz, J., Schrag, D., Hampt, G., Murray, D.W., 1996. Pore fluid constraints on
Genthon, C., Kotlyakov, V.M., Lipenkov, V., Lorius, C., Petit, J.R., the temperature and oxygen isotopic composition of the glacial
Raynaud, D., Raisbeck, G., Ritz, C., Sowers, T., Stievenard, M., ocean. Science 272, 1930–1932.You can also read
